Higher Risk of Loosening for a Four-Pegged TKA Tibial Baseplate Than for a Stemmed One: A Register-based Study

Abstract

Background

Modern modular implants allow surgeons to mix different combinations of components within the same brand. From 1999 to 2012, the NexGen®-CR Option femoral component, together with a NexGen® Option Stemmed tibial plate (stemmed baseplate), which uses a short central stem, was the most-frequently used NexGen® combination in the Swedish Knee Arthroplasty Register. However, from 1999 to 2012, the same femoral component was also used along with the NexGen® Precoat four-pegged tibial baseplate (pegged baseplate). Considering the difference in the fixation concepts for these two tibial baseplates, we wanted to study whether their revision rates differed.

Questions/purposes

To investigate the difference in (1) all-cause revision and (2) the risk of revision for aseptic loosening between the NexGen® pegged and stemmed baseplates when used with the NG-CR Option femoral component and the same two types of inserts.

Methods

The Swedish Knee Arthroplasty Register provided data. The register, which was started in 1975, has since 1999 registered part numbers for individual implant components, allowing it to assess the combinations of components used in each patient. It has been shown to have high completeness (97%) and validity [12, 15]. The inclusion period was 1999 to 2012; during that time, 137,143 primary knee arthroplasties were registered, of which 125,094 were TKAs. Only TKAs performed for osteoarthritis and without patellar resurfacing were included, since not resurfacing the patella is the standard procedure in Sweden. This left 15,287 knees with the stemmed baseplate and 2479 with the pegged baseplate, or 12% and 2% of the total number of TKAs, respectively. Two general hospitals used the pegged baseplate exclusively during that period. Thus, specific patients were not selected for having the pegged plate. The mean age, mortality, and length of followup were similar for the two groups.

We used the Kaplan-Meier statistics to calculate the cumulative revision rate (CRR) and Cox regression to compare risk ratios after adjusting for age and sex. The end point was a knee revision for respective all causes or aseptic loosening. The study ended on December 31, 2016. Due to the free healthcare system in Sweden it is highly unusual for patients to seek elective revision abroad, and by use of the extensive Swedish census register, we estimate the level of followup approximately 97%.

Results

Knees with the pegged baseplate had a higher risk for all-cause revision than did those with the stemmed baseplate (5.8% [95% confidence interval {CI}, 4–8.3] and 3% [95% CI, 2.6–3.5] at 15 years; p = 0.003). After controlling for age and sex, the aseptic loosening risk in the pegged baseplate group was still higher than that in the stemmed group (relative risk, 5.40; 95% CI, 3.64–8.02; p < 0.001).

Conclusions

In this Swedish registry study, we observed a higher loosening risk with the pegged baseplate than the stemmed one, even after controlling for age and sex. Because this was only a comparison of implants from one vendor, and because there may have been other between-group differences for which we could not fully control, this concerning finding should be explored using data from other registries.

Level of Evidence

Level III, therapeutic study.

Introduction

The effect of tibial component stem design on load distribution and fixation has been of interest for a long time [17, 18]. However, in spite of theoretical advantages and disadvantages for using a stem or not, there are no evidence-based recommendations that guide this choice for primary TKAs. Since the introduction of the NexGen® (Zimmer, Warsaw, IN, USA) TKA, both a four-pegged baseplate as well as baseplate with a short central stem have been available for cemented use. So far, a few small studies have evaluated the NexGen® pegged baseplate [2, 3, 13], and to our knowledge, only one small study [3] compared the pegged version of this implant to the stemmed version. Bertin et al. [3] in 2007 found no difference between the devices (98% survival for both implants at 7 years), but reported on only 225 knees, which is small to compare the endpoint of revision. A single surgeon series of 87 patients with the pegged baseplate, followed for 10 to 12 years [2], reported 97% 10-year survival while Schwartz et al. [13] reported 98% survival for the pegged baseplate at 10-years in a series of 179 knees. Since those studies evaluated no more than 225 knees each, larger populations are needed, and national registries provide a robust mechanism to make such comparisons across large numbers of patients. Because both the stemmed and pegged version of this implant remain in service, doing so seems very important.

In arthroplasty registries, the comparison of different designs is complicated by the modularity of modern prostheses allowing surgeons to combine components, for example, the same baseplate with different types of femur components and/or inserts, which themselves may affect the outcome. In 1999, the Swedish Knee Arthroplasty Register began recording the part numbers (catalog numbers) of all inserted components, allowing it to assess what combinations of components are used in each TKA. During classification of the part number series for the NexGen implants in the register, we became aware that many cemented NexGen pegged baseplates had been used during 1999 to 2012, in combination with the same femoral components and inserts as a larger number of the more commonly used cemented NexGen stemmed baseplate. Unfortunately, we cannot estimate the worldwide usage of the NexGen four-pegged plate, nor can we find it mentioned in reports of other national arthroplasty registers. However, our relatively large register dataset on the two baseplates with a followup for up to 17 years, spurred us to explore whether differences in the revision rate between the designs could be detected. Although the comparison is implant/vendor-specific and only concerns cemented implants, we felt it would be a relevant contribution into the discussion on the effect of stems versus pegs on long-term fixation.

We therefore sought to determine the difference in (1) all-cause revision and (2) the risk of revision for aseptic loosening, between the NexGen® pegged and stemmed tibial baseplates, when used with the NexGen-CR Option femoral component and the same two types of inserts.

Patients and Methods

The study is a register study from the Swedish Knee Arthroplasty register. It is the oldest national arthroplasty register, established in 1975, and has been shown to have a high degree of completeness and validity [12, 15].

The study concerns the period from 1999, when the register started registering part numbers to the end of 2012, when use of the NexGen pegged baseplate ceased in Sweden.

During this period, 137,143 primary knee arthroplasties were performed in Sweden and entered into the Swedish Knee Arthroplasty Register (Fig. 1). Of these, 125,094 were TKAs (91%) of which 36,810 (29%) were of the NexGen brand. Of these, 854 (2%) were excluded because one or more part numbers for the femur, tibia, or insert were missing. This left 35,956 NexGen TKAs, using more than 30 different combinations of femoral components and tibial baseplates; the most common combination was the NexGen-CR Option femoral component with the stemmed tibial baseplate, used in 16,185 knees (45%) while the same femoral component was used with a four-pegged baseplate in 2683 knees (7.5%).

Fig. 1.

A STROBE flow chart breaks down the study population.

To make the comparison of the two plates more uniform, we only included knees operated on for osteoarthritis, by excluding 776 TKAs performed for other diagnoses (630 [3.9%] in the stemmed baseplate group and 146 [5.4%] in the pegged group).

We considered only TKAs using two types of regular constraint inserts (the NexGen -CR/CRA and the NexGen -CR/CR-flex/CRA). Zimmer introduced the latter insert in 2005, which completely replaced the CR/CRA (cruciate-retaining/cruciate-retaining augmentable) version in both implant groups within a few years (Table 1). Thus, 156 TKAs having other types of inserts were excluded (102 [0.7%] in the stemmed baseplate group and 54 [2.1%] in the pegged group).

Table 1.

Number of the two types of components that were inserted from 1999 to 2012

Finally, 170 knees undergoing patellar resurfacing were excluded (166 [1%] in the stemmed baseplate group and four [0.2%] in the pegged group), but resurfacing of the patella in TKA is not often performed in Sweden. This left us with 15,287 TKAs in the stemmed group and 2479 TKAs in the pegged group (Fig. 1). Although we excluded TKAs not inserted for osteoarthritis, such as those with patellar resurfacing and those not using one of the two types of regular constraint inserts, we also checked if not excluding them changed the results. We found that the risk ratios were slightly affected but otherwise it did not change our findings (data not shown).

When the NexGen-CR Option TKA was introduced in Sweden, both stemmed and pegged baseplates for cemented use were available, the latter similar in design to the Miller-Galante TKA baseplate. In all, 47 Swedish hospitals selected the NexGen stemmed baseplate, but two hospitals decided to exclusively use the NexGen pegged baseplate, probably because they had used the Miller-Galante pegged baseplate before starting to use NexGen. The two were general hospitals (not specialized referral centers) serving the general population in the region, and they used the pegged baseplate exclusively for primary TKAs during the period. However, in late 2012, they switched from the pegged to the stemmed version, mainly because of lobbying by the implant provider. Four additional hospitals tested the pegged baseplate but only performed one to two surgeries each.

The mean followup period for the stemmed baseplate group was 8.1 years (range, 0–18.0 years) and 8.4 years in the pegged group (range, 0–17.1 years). This must be considered a relatively small difference with little effect on our survival calculations.

The difference in the distribution of age and sex was not clinically relevant between the groups (Table 2). However, being unable to adjust for other patient related factors, such as height, weight, smoking or social status, the similarity of the two patient groups at baseline was a concern. Thus, we investigated if the overall mortality between the groups was different and found no difference (RR, 1.02 [95% CI, 0.94–1.11].

Table 2.

Patient demographics for each group

Statistical Analyses

Our primary study outcome variable was the relative risk of revision from all-causes between the two baseplate groups during the analysis period and the secondary outcome variable was the relative risk of revision for only aseptic loosening. For descriptive statistics, we used means and proportions. We calculated the cumulative revision rate curves using the Kaplan-Meier method using 1-month intervals. The log-rank test was used to test for equality of the survival for the two baseplate over the period. The 95% confidence intervals (CIs) were calculated using the Wilson quadratic equation with Greenwood and Peto effective sample size estimates [5]. The curves were cut off when 40 knees remained at risk. We used a Cox regression analysis to compare the relative risk for revision between the groups as well as for mortality, while adjusting for differences in age and sex. The proportional hazards assumption of the Cox regression model was assessed by visual inspection (log-minus-log plot). Any p value < 0.05 was considered significant.

The statistical software used was: STATA: Release 15 (StataCorp LLC. College Station, TX, USA).

Results

All-cause Revision

The Kaplan-Meier analysis showed that patients in the pegged group had a higher risk of revision for any reason than did those in the stemmed group (log-rank test; p = 0.003). At 15 years, the cumulative risk was 5.8% [95% CI, 3.4–11.4] and 3% [95% CI, 2.6–5.5] respectively (Fig. 2).

Fig. 2.

This figure shows the cumulative revision rate (CRR) for all causes. The red solid line shows the CRR for the pegged plate and the blue line shows the CRR for the stemmed plate. The shaded areas show the 95% CI.

After adjusting for age and sex, the Cox regression analysis showed that the relative risk of revision for the pegged group was higher than for the stemmed group (RR, 1.46 [95% CI, 1.14–1.85]; p = 0.002). Fifty-two of 338 revisions in the stemmed group and 47 of 81 in the pegged group were performed for loosening (Table 3).

Table 3.

Number of revisions and reasons for revisions in each group

Revision for Aseptic Loosening

When revision for only aseptic loosening was considered the endpoint, the difference favoring the stemmed group was even larger (Fig. 3). At 15 years, the cumulative risk was 3.9% (95% CI, 2.3–8.8) and 0.6% (95% CI, 0.4–2.2), respectively. The Cox regression analysis after adjusting for age and sex showed that the relative risk of revision for aseptic loosening for the pegged group as compared to the stemmed group was still higher (RR, 5.40 [95% CI, 3.64–8.02]; p < 0.001).

Fig. 3.

The cumulative revision rate (CRR) for aseptic loosening only is shown. The red solid line shows the CRR for the pegged plate and the blue line shows the CRR for the stemmed plate. The shaded areas show the 95% CI.

The relative risk of revision, for other causes than loosening, was not different between the groups (data not shown).

Discussion

There are many different combinations and configurations of components that can be used in TKA. The modularity of modern implants often makes it possible to combine different components within the same brand with the result that TKAs with a certain brand name may in fact represent TKAs having quite different properties. For the arthroplasty registries, modularity complicates classifications and reduces the sizes of groups that are comparable, which might help explain why other registers have not reported results separately for the pegged plate. That we managed to identify a reasonable number of the two baseplates being used with the same femoral components and polyethylene inserts makes this register study interesting as the only implant difference between the groups was the baseplate used. To our knowledge, only one small study has compared the risk of revision between cemented pegged and stemmed plates [3] but that study was too small to identify differences that could be clinically important. For this reason, we felt it important to explore the question in the setting of a large national registry.

This study had several limitations. We note that the pegged plates were used at only two hospitals. However, the surgeons at general hospitals used these implants exclusively, and did not select different implants for different patients. It is still possible that the local population that the two hospitals were serving was different from that of the other hospitals with respect to health, weight, smoking, physical activity or social status. We were not able to adjust for these factors, but age, sex, and mortality were similar in the two groups, and we found no evidence for major differences between the two groups. Similarly, we think it unlikely that there are discrepancies in surgical practice or skill which could explain our findings. We note that revision is a blunt measure of outcome. As many register studies using revision as an endpoint, this study lacks both clinical and radiological followup. However, revision as an outcome measure is an ultimate proof of that both the patient and the surgeon agreed on that the primary arthroplasty had problem(s) warranting surgical intervention. Thus, we consider it relevant. Finally, this study concerns two vendor-specific cemented baseplates and the results do not necessarily apply for other cemented stemmed or pegged baseplates, or for uncemented baseplates that rely on bone ingrowth instead of cement for fixation.

Despite the theoretical advantages of using small pegs instead of short stems, the clinical advantage seems dubious, and our results suggest that the pegged design may in fact be associated with a greater revision risk. Although the first modern condylar TKA design, which was introduced in 1973, had a short tibial stem, the benefit of this design is still being debated [14, 16]. One reason not to use a short central stem (50 mm or less) is that it may cause stress shielding, which may lead to loosening and subsequent revisions [8, 9]. However, this has not been verified in other studies. On the contrary, national arthroplasty registers have shown favorable short- and long-term prosthetic survival for several types of implants with cemented, short, central stems, indicating that proximal stress shielding is of limited clinical relevance [1, 15]. Other reasons for having a tibial baseplate with pegs instead of a stem have been preservation of the host bone, compatibility with minimally invasive surgery, and easier removal with revision techniques [2]. However, regarding preservation of the host bone, the relevance of displacement of the central cancellous bone by a short stem is questionable. In addition, while the removal of a baseplate with only pegs may be little easier because the baseplate does not have to be lifted proximally, this advantage diminishes if the pegged plate is replaced by a stemmed one. It has also been suggested that a stemmed baseplate causes increased pressure within the intramedullary canal and thus may increase the risk of fat embolism. However, the risk is low and may also be related to insertion of intramedullary guides rather than to preparation and insertion of a short stem [7, 10].

We are only aware of one midterm study directly comparing the stemmed and pegged NexGen tibia baseplates. Analyzing 84 stemmed and 141 pegged TKAs after a 7-year followup period, Bertin [3] suggested that pegged tibial components offer comparable results to stemmed components, with both groups having good clinical results, no loosening or radiographic evidence of loosening, and a 98% survival rate. The 2% all-cause revision proportion was not far from what we observed in our study at 7 years (Fig 2). However, due to a smaller material and shorter followup, Bertin [3] noted only one revision in the stemmed group and two revisions in the pegged group. Thus, it lacked the power of our study to detect differences between the groups. Other studies have reported results only for the pegged plate. In a single surgeon series of 87 patients, Barrington et al. [2] reported a 97% 10-year survival with one patient revised for aseptic loosening. Schwartz et al. [13] reported 98% survival at 10 years in a four-surgeon series of 179 knees, with no revisions for aseptic loosening.

With respect to revisions for loosening, these accounted for less than a quarter of all the revisions in our material (Table 3). Considering that the aforementioned studies had a shorter followup than our study, their finding of one revision for loosening among a total of eight revisions is probably not so peculiar.

Our study showed a higher revision rate for knees with the pegged baseplate than for knees with the stemmed baseplate, mainly because of loosening. It is known that in addition to shear forces, there are varus and valgus compression and tension forces, as well as roll back and rotation forces that affect pegged and stemmed plates differently, and this may possibly explain the observed differences [7, 19].

Although we cannot find support for our result in the current evidence, we do not consider it being at odds with it. On the contrary, it is reasonable to attribute our finding to a larger cohort size and thus a greater power. However, our results need to be confirmed by others, but due to the low revision rate for the current TKA implants, this calls either for larger cohort size, such that are currently mainly available in registries, or different methods for evaluating the quality of the fixation before the loosening actually occurs.

In conclusion, we observed a higher risk of loosening with the pegged tibial baseplate than the stemmed one, after controlling for age and sex. Because this was only a comparison of implants from one vendor, and because there may have been other between-group differences that we could not fully control for, this concerning finding should be explored using data from other registries.